What is a t-value in trading? In plain words

EA Trade Tester Demo lab How we test All guides For AI agents

Published 12 November 2026 · methods v2.5 · by TrueTest

The t-value answers one question about a record: is the average trade far enough from zero that chance alone is an unlikely explanation? It is the number our verdicts are built on. What it cannot do: it says nothing about the future, it cannot see faults in the code or costs that are missing from the record, and on its own it does not know how many versions you tried before this one.

See the t-value of my EA See a full example first

Drop your .mq5/.mq4 code, a Python bot or an MT5 report on the home page. The first read takes seconds and needs no sign-up; the lab verdict is free too. Only the full report is paid (GBP 49: a PDF to keep plus the interactive report). The price shown is the total price: no VAT is added. TrueTest is not VAT-registered in the UK. We read your code as text and never run it as given; an outside AI service drafts the rule card only if you tick the box (see Security). Not financial advice.

How is a t-value worked out?

Take the average result of a trade, in R (multiples of the risk) or in money. Then work out how much that average could wobble by chance: the spread of the trade results (the standard deviation) divided by the square root of the number of trades. The t-value is the average divided by that wobble. The further it is from zero, the less likely the average is an accident.

A worked example

A record has 150 trades, an average of +0.12 R and a standard deviation of 1.1 R. Its t-value is 0.12 / (1.1 / √150) = 1.34. That is below 1.96, the usual bar for one planned test, so the average cannot be told apart from zero: NO EVIDENCE.

The same average and spread over 600 trades gives 0.12 / (1.1 / √600) = 2.67. Four times the trades doubles the t-value. For one planned test that clears 1.96. If it was the best of 100 tries, the luck bar is 2.53, and 2.67 clears it.

Why are trades on the same day grouped?

Our t-value is the t-value of the average trade with trades on the same day grouped together (day-clustered), because trades on one day are not independent. Five trades opened on the same news spike are closer to one bet than to five. Counting them as five would make the wobble look smaller and the t-value bigger than the record deserves, so we add up each day first and measure the wobble of the days.

When are there too few trades?

Below 30 trades or 20 trading days the verdict is TOO FEW: the wobble itself cannot be measured well enough, so even a large t-value means little.

How does the t-value meet the luck bar?

The luck bar is the t-value the best of your tries would reach with no edge at all. It grows with the number of versions you tried, and a result has to clear it to count. The table and the maths are on real or luck; you can try your own numbers in the luck bar calculator.

Questions people ask

Is a higher t-value always better?

A higher t-value means the average is less likely to be chance. A t-value far above what the strategy type usually shows is a reason to look for look-ahead or missing costs.

Do I need R values to work it out?

No. Money per trade works too, at a fixed lot size; the t-value does not depend on the unit.

Is a t-value the same as a win rate?

No. A 90% win rate with rare large losses can have a t-value near zero.

Is this financial advice?

No. It explains a statistic. Not financial advice.